. You might lose your job because of AI. Maybe you already did. Maybe you are struggling to enter an industry that now hires fewer juniors. And if your job survives, competition may become harsher: fewer openings, higher expectations, lower bargaining power.
However, AI alone is not a danger. It is AI deployed inside an economy where workers have little leverage over how productivity gains are distributed.
AI Anxiety Is Rational
Technological change has always reshaped labor markets. Some jobs disappear, others emerge, and society adapts. So isn’t AI just another technology that will make some jobs obsolete in favor of more efficiency while creating new jobs? Many argue it is not. While it is hard to have a clear vision of what the future will look like, the belief is that AI is fundamentally different because of the scale and speed of the disruption it could cause.
New agentic models can perform some programming tasks at a level approaching or exceeding experienced developers, and even show human qualities such as judgment and taste [1]. Software engineering is only the first of many knowledge industries being transformed, as an AI which can code “can help build the next version of itself”.
However, the impact of AI is visible in many different occupations. A translator who specialized in routine business documents sees clients disappear almost overnight. A customer support worker is told to supervise chatbots instead of talking to people. A junior legal assistant who once learned by summarizing case law or doing legal research now competes with seniors augmented by AI tools.
Some are worried that a large share of computer-based knowledge work could eventually be automated, with massive layoffs happening simulatenously [2]. Even though AI might create some new jobs, it doesn’t look like the new jobs created by AI will compensate for the scale and quality of jobs being displaced.
Worse, job displacement is not merely a side effect. Many companies explicitly view AI as a tool for reducing headcount. We see this from the idealization of “a single-person-company […] reaching a billion dollar valuation” [3] and the controversial ad campaigns inciting to “stop hiring humans” in favor of AI, which “won’t complain about work-life balance” [4].
AI Automation Is Not The Real Problem
Businesses operating in competitive markets are structurally incentivized to reduce costs and increase productivity. A relevant part of it is to reduce the costs related to labor. So, if a technology exists, that is cheaper and more efficient than human labor, it is in the company’s interest to replace or reduce the number of humans involved in production. Costs go down, dividends go up. Nobody would argue against this picture. But this vision makes technology good for company owners, and bad for workers, promoting anti-technology stances such as Luddism.
There is a surprisingly overlooked way technology could benefit workers.
Imagine a worker who becomes ten times more productive because of AI. Let’s say that production needs to remain unchanged due to unchanged demand. Society could respond in at least two ways:
- One worker keeps the job while nine lose theirs.
- Ten workers keep their jobs, but all work far fewer hours.
Technology alone does not determine which outcome happens.
Politics, labor power, and economic institutions do.
Increased Productivity Did Not Decrease Working Hours
History suggests that productivity gains are rarely distributed automatically in favor of workers. For example, since the most valuable thing we have in life is time, and technology is requiring less time to perform tasks, we would expect this to reflect on working hours. Since World War II, workers became dramatically more productive thanks to industrialization, computers, logistics, and automation.
Yet, most productivity gains were not converted into leisure time.² They were converted into profits, executive compensation, and shareholder value.
Of course, sharing productivity gains does not necessarily mean only shorter working hours. It can also mean higher wages, stronger public services, worker ownership, or broader social participation in the wealth generated by automation. But if the overall impact of a technology is increasing inequality, we know that the gains were not shared enough.
So the key political and economic question is how the resulting gains are distributed. If they go too much in favor of the owners, it might cause the catastrophic effects that we described earlier. If it is used in favor of the workers, it can improve the lives of ordinary people. Major historical economists, from Mill to Keynes, had a similar utopian point of view, arguing that the ultimate purpose of technology would be to reduce the amount of time devoted to work [5, 6].
Productivity Gains Have Always Been Political
So the main point is how we are able to deal with this tension between the benefit of investors rather than workers. During the Industrial Revolution,¹ workers did not automatically benefit from productivity gains. Factory owners became enormously wealthy while laborers endured brutal conditions and long hours [7]. Improvements came later — through unions, political pressure, labor laws, and collective bargaining.
Since the 1980s, labor movements became weaker, and the general trend has been that the impact of technology brought much more benefit to owners than to workers [8]. Labor movements are now so weak, that we take for granted that all the benefit related to technology should go to company owners.
AI Could Create Two Futures
If the benefits of AI are skewed towards benefiting investors (as they were in the past 50 years), the result could be severe job displacement, rising homelessness, and extreme inequality. Economic power could become concentrated among a tiny number of firms and investors, while large parts of the population face growing precariousness. Even proposals like universal basic income [2] ultimately raise the same question: _who owns the wealth generated by automation?_³
If we manage to distribute the benefits of AI in favor of workers, technological progress will not necessarily translate into mass insecurity. People will be able to work less instead of becoming disposable. Working fewer hours would leave more time for creativity, relationships, health, civic participation, and rest.
So, the question is not “Will AI replace us?”, but rather:
“Who captures AI’s productivity gains”?
What You and I Should Fight For
So, AI is not stealing your job. With AI, you can keep your job while having a better life. But this requires acknowledging the relations of power existing with labor. If there are no mechanisms to manage the benefits of AI in favor of the community at large, AI will drastically increase inequality. This might require fighting to defend your rights, but if you choose the wrong target, you will fail.
Your target is inequality, not technology.
Of course, businesses cannot simply reduce working hours overnight while keeping salaries unchanged. Competition, international markets, and profit pressures are real constraints. But that is precisely the point: the economic and social system determines how technological gains are distributed. If the economic and societal pressures only reward profit maximization, AI will not improve labor conditions. It will intensify existing inequalities instead.
If the interest of the labor force is instead taken into account, we can hope for the optimal sweet spot between production and well-being to be reached. This can be achieved by fighting for workers’ rights and better regulation, but there may be many ways. Worker cooperatives are one example of an economic structure where workers directly benefit from productivity gains, and publicly owned AI — as it is for example being pursued in Switzerland [9]— is another.
The Real Question
AI is not forcing your employer to fire you. This is an active choice. Every major technological revolution raises the same political question:
Who benefits?
If AI’s gains are captured exclusively by owners and shareholders, inequality will deepen dramatically.
If gains are shared broadly, AI could reduce working hours, improve quality of life, and expand human freedom on a scale previous generations only imagined.
AI will not decide which future we get.
We will.
References
[1] Matt Schumer, Something Big Is Happening, 2026. https://x.com/mattshumer_/status/2021256989876109403
[2] A. Yang, The End of the Office, 2026, https://blog.andrewyang.com/p/the-end-of-the-office
[3] A. Ohanian, https://x.com/alexisohanian/status/1752753792058294725
[4] M.A. Bitoon, “Stop hiring humans” billboards provoke backlash against AI amid tech layoffs. https://www.greenbot.com/stop-hiring-humans-billboards/
[5] J.S. Mill, Principles of Political Economy, 1848.
[6] Keynes, J. M. (1930). Essays in persuasion. Palgrave Macmillan London. https://doi.org/10.1007/978-1-349-59072-8
[7] E.J. Hobsbawm. “The standard of living during the Industrial Revolution: a discussion.” The Economic History Review 16.1 (1963): 119–134.
[8] The Productivity–Pay Gap, Economic Policy Institute, March 2026 https://www.epi.org/productivity-pay-gap/
[9] https://apertvs.ai/; https://ethz.ch/en/news-and-events/eth-news/news/2025/09/press-release-apertus-a-fully-open-transparent-multilingual-language-model.html
[10] The cited data can be obtained through the ourworldindata database: https://ourworldindata.org/grapher/labor-productivity-per-hour-pennworldtable?tab=line&country=USADEUKOR~GRC. Those statistics are obtained from Feenstra et al. — Penn World Table (2025)
Footnotes
- Things went very differently in many parts of the world, for example between colonizer and colonized states, different kinds of legislation, levels of democracy, and of economy. I still prefer to remain generic, since being more specific would lead us away from the main point of the text, which is AI.
- Data from the World Penn Table shows that, since 1950, the average productivity has increased by roughly a factor 4 in the USA, 10 in Greece, 15 in Germany, and 30 in South Korea [10]. However, working hours have not decreased by a similar factor. With the exception of few European countries, they haven’t decreased at all.
- In a society where labor is replaced by AI, labor taxes would be very low, so to healthily support a universal basic income, a switch from a labor to a wealth taxation would be necessary.
Images in this article
All images where generated by the author using chatGPT5.5.
Instagram Engineering
Shopify Engineering